Industrial robots can be used to move an object in space according to a certain number of degrees of freedom of the latter. These robots are used for moving items which are typically too heavy or too bulky to be moved by a human operator, on production lines for example. These robots can also provide precise positioning of these items, for the purpose of assembly operations, for example. Not all the tasks performed by industrial robots can be entirely automated, and some of them must therefore be controlled by a human operator. Typically, an industrial robot is a system comprising a plurality of joints, made after the pattern of a human arm. It may take the form of a manipulator arm, fitted at one end with a gripping member adapted to grasp the item. The item may be oriented spatially by the robot, for example by rotation about three axes, and by translation along the said three axes; in all cases, the combined movements of the components of the robot must allow the item to be manipulated so as to be moved and oriented in space.
There are various systems for controlling these robots. In a first technique, commonly called “remote operation”, an operator is able to control the robot at a distance, by means of an interaction referred to as indirect. Typically, the robot may be controlled by means of a control device which conventionally takes the form of a control box having a plurality of push buttons for initiating various movement actions. The control box can then be used to move the gripping member of the robot, pressure on a button being, for example, associated with a movement according to one of the six degrees of freedom, in a given direction.
A drawback of this technique is that a control box does not allow the operator to sense the forces applied by the robot and the environment to the manipulated item, although this data feedback may be essential for the correct performance of certain tasks such as assembly tasks.
A further drawback of this technique is that it allows the movement of the item to be controlled only within frames of reference associated with the robot, or more precisely with the robot's gripping member, or with one of the articulations of the manipulator arm, for example. However, the operator is concerned with the manipulation of the item itself, rather than with the control of the robot as such: he may therefore prefer to be able to control the movement of the item in different frames of reference, such as those associated with the item. Thus a remote operation technique may be useful for placing the robot in a particular configuration in space, but it is less suitable for the precise manipulation of an item, the required degree of precision being, for example, greater when it is necessary for the item to be in contact with its environment.
An alternative known technique, commonly referred to as “computer-assisted remote operation with force feedback” can partially overcome the aforesaid drawbacks. According to this technique, the control device is not made in the form of a box fitted with buttons, but rather in the form of a member which may be called a “master arm”, for example a control lever, or “joystick” as it is called in English, which can be moved in space by the operator, and causes the movement of the robot, which is then called a “slave”. The control lever may be provided with movements controlled by a dedicated controller, allowing the provision of a force feedback perceptible to the user. One advantage of this technique is that it allows more intuitive control of the robot. Furthermore, the fact that this technique allows the operator to sense the forces applied to the robot helps to provide more precise assistance to the latter, notably during the performance of tasks in which the manipulated item comes into contact with external elements. According to this technique, the operator is able to choose frames of reference in which the robot is to be controlled to perform specific tasks; indeed, the operator may even be able to specify the tasks to be performed by the robot.
A drawback of this technique is that it requires the operator to specify the frames of reference or the tasks, for example if he wishes to manipulate the item in frames of reference other than those of the robot, which may prove to be difficult and, in particular, impractical during the manipulation of an item.
A further drawback of the two aforesaid techniques relates to the operator's viewpoint of the scene, in that the operator has to remain at a specific location, and is not free to choose his viewpoint of the scene, unless he uses video camera systems, for example; that is to say, unless he accepts additional constraints in terms of the number of sensors used, as well as constraints relating to the positioning of these sensors, notably for the purpose of avoiding possible obscuration.
It may thus be preferable for the operator to remain in the proximity of the scene. There are known control systems by means of which the operator can interact directly with the robot, for example by using a control handle fixed to the gripping member of the robot, and enabling the operator to cause the movement of this member. In this way, the item can be manipulated with its six degrees of freedom, with the provision of exact compensation for the weight of the item. However, more especially if the operator has to manipulate a large item in a precise manner, he may find it difficult to control the robot simply by means of a control handle fixed to one end of the robot's arm. It may be essential for the operator to grasp a particular location on the item in order to manipulate it in a certain way; moreover, the handle fixed on the robot may move out of the operator's reach when the robot has to grasp a large item on its own.
In order to overcome these drawbacks, control systems have been designed which enable an operator to interact with an item to be manipulated by means of the item itself, by using a control member, separate from the robot and allowing direct interaction with the item, located at one end of the item. A control system of this type is described in the Japanese patent application published under the reference JP 2008/213119. In this control system, a handle separate from the robot may be positioned at a predetermined location on a frame supporting the item to be manipulated.
A drawback of this control system arises from the fact that the item can be controlled only in the frame of reference of the robot or of the handle, which is fixed with respect to the control system, making it necessary for the orientation and position of the frame of reference of the handle with respect to the robot to be predetermined and known during the implementation and adjustment of the robot control, and to remain fixed subsequently.
Furthermore, the grasping position may not always be suitable for the manipulation of the item: for example, if the item has to be turned over, the handle may be located under the item when the latter has been turned over, causing difficulties in manipulation. The grasping position may also prove to be unsuitable if, for example, the item has to be assembled with another item, as the handle may obstruct the assembly operation in some configurations.
A further drawback of systems in which the forces applied by the human operator are applied remotely from the robot is due to the fact that the forces applied by the operator at the point where he grasps the item are different from the forces experienced by the gripping member of the robot, possibly giving rise to ambiguities between different movements, for example rotational and translational movements. For example, in a configuration in which the operator and the gripping member of the robot grasp each end of an item, for example a plank, the force experienced by the gripping member of the robot resembles a torque, regardless of whether the operator wishes to make the item pivot about the robot's effector or to translate it.